Methods in Bioengineering: Biomicrofabrication and … · 2011-11-16 · nant kinesin, purifying and ... >100,000 x g is good, but less is acceptable) to pellet insoluble material

C H A P T E R

1 3Microtubule Motors in Microfluidics

Maruti Uppalapati,1 Ying-Ming Huang,2 Shankar Shastry,1 Thomas N. Jackson,2 andWilliam O. Hancock1*1 Department of Bioengineering and 2 Department of Electrical EngineeringPenn State University, University Park, PA 16802

* Corresponding author: William O. Hancock, 205 Hallowell Bldg., University Park, PA 16802, phone:814-863-0492, fax: 814-863-0490, e-mail: [email protected]

311

Abstract

Kinesin motor proteins carry out a range of transport functions in eukaryoticcells including long-distance transport in neurons and the movement ofchromosomes during cell division. Because emerging microfluidic devices uti-lize channel geometries similar to cellular scales and because these devicesgenerally require transport through these channels, there is significantinterest in incorporating biomotor-driven transport into microfluidic devices.Kinesin-driven transport has the advantage of being able to work againstconcentration gradients and bulk fluid flow, and microtubules can befunctionalized to carry a range of cargo. This chapter describes the founda-tional methods for generating kinesin-driven transport in engineeredmicrochannels. Protocols are described for expressing and purifying recombi-nant kinesin, purifying and fluorescently labeling bovine brain tubulin, andcarrying out the microtubule gliding assay. General design considerations arepresented for integrating kinesins and microtubules into microchannels,followed by a detailed description of a microfluidic device prototype thatsuccessfully integrates microtubule-based transport.

Key terms kinesinmicrotubuletransportmotor proteinbionanotechnology

13.1 Introduction

In emerging lab-on-a-chip technologies, the microfluidic devices designed to shrinkanalytical tests and biosensors down to microscale dimensions generally require somemanner of mechanical actuation. While these mechanical forces can be achievedby pressure-driven fluid pumps, electrically or magnetically driven movement,electroosmotic flow, or other approaches, there is a continuing need in this area forreliable mechanical actuation and directed transport. In eukaryotic cells, kinesin motorproteins transport intracellular cargo and provide the mechanical forces underlyingchromosome movements and other mechanical processes. These motor proteinsutilize the chemical energy of ATP hydrolysis to walk on protein filaments, calledmicrotubules. This biologically driven movement can be reconstituted under a micro-scope by immobilizing kinesin motors on a substrate (such as a glass coverslip) andobserving the movement of microtubules over the surface. Building on the robustmicroscale transport in this “microtubule gliding assay,” there has been considerableinterest in utilizing kinesin-driven microtubule motion for nano/microscale transportapplications in lab-on-a-chip devices [1–3]. While the development of lab-on-a-chipdevices is primarily based on microfluidics technology, integration of kinesin-drivenmotion in microfluidic devices has several advantages. First, motors can transport cargoagainst concentration gradients. Second, no bulk fluid flow or pumping mechanismsare required, easing the fabrication and power requirements of such devices. Third,microtubules and kinesins can be modified to bind specific cargo, enabling their use inbiosensing and separation processes.

A schematic of a potential device of this type is illustrated in Figure 13.1. This deviceis designed to detect viral RNA in biosensing applications, or to detect mRNA from a celllysate, which would enable gene expression profiling, in principle down to single celllevels [1]. In the design, microtubules functionalized with single-stranded DNA con-structs (molecular beacons) are transported through microchannels functionalized withkinesin motors [4]. These mobile probes capture and transport specific RNA and reportthis sequence-specific binding by unquenching their fluorescence. Integrated opticalsensors and electrodes detect and guide the microtubules such that the cargo is sepa-rated from the heterogeneous sample and is concentrated in the collection chamber.If successful, this device could have a range of uses, and it is only one of manybiomotor-driven devices that have been envisioned. However, successful integration ofmicrotubules in such lab-on-a-chip devices requires (1) the ability to achieve directionalmotion of microtubules in enclosed microfluidic channels, (2) the ability to bind andtransport cargo, and (3) the ability to detect and sort microtubules based on the presenceof cargo. This is an active area of research and key blocks of technology have been devel-oped by our group and others. In this chapter we detail the experimental methodsinvolved in purifying kinesin motors and microtubules, techniques for modifying theseproteins for use in microfluidic channels, and approaches for optimizing the design ofmicrofluidic channels to achieve robust kinesin-driven transport.

13.2 Materials

Relevant materials, reagents, and equipment are listed below. Both the “Materials” andthe “Methods” sections are divided into four subsections: (1) Kinesin expression and

Microtubule Motors in Microfluidics

312

purification, (2) tubulin purification and labeling, (3) microtubule gliding assay, and (4)integrating motility into microchannels. Unless specified, the reagents mentionedbelow can be obtained from alternative vendors, and for reagents where no source isspecified, Sigma, Fischer, or any equivalent source are sufficient.

13.2.1 Kinesin expression and purification materials

13.2.1.1 Kinesin reagents

• Kinesin expression plasmid in BL21(DE3) bacterial cells

• LB media (Qbiogene, Cat # 3002-092)

• Ampicillin sodium salt (Sigma-Aldrich, Cat # A9518)

• Isopropyl β-D-1-thiogalactopyranoside (IPTG) (Sigma-Aldrich, Cat # I5502)

• Phenylmethanesulfonyl fluoride (PMSF) (Sigma-Aldrich, Cat # P7626)

• Adenosine-5’-triphosphate (ATP) (Plenum Scientific Research, Inc., Cat # 0163-04)

• ß-Mercaptoethanol (ßME) (Sigma-Aldrich, Cat # M3148)

• Dithiothreitol (DTT) (Bio-Rad, Cat # 161-0611)

• Sodium phosphate monobasic (Sigma-Aldrich, Cat # S3139)

• Sodium phosphate dibasic (Sigma-Aldrich, Cat # S3264)

• Sodium chloride (Sigma-Aldrich, Cat # S3014)

• Magnesium chloride (Sigma-Aldrich, Cat # 68475)

• Sodium hydroxide (NaOH)

• Sucrose (Sigma-Aldrich, Cat # S0389)

• Imidazole (Sigma-Aldrich, Cat # I5513)

13.2 Materials

313

Figure 13.1 Schematic for biosensor/bioseparation device-based kinesin-driven transport.Microtubules are functionalized with molecular beacons that bind RNA and fluoresce upon binding.The beacon-functionalized microtubules are transported through microfluidic channels to whichkinesin motor proteins are adsorbed. An integrated photodiode (PD) and organic light emitting diode(OLED) detect the presence of RNA cargo on the microtubule and electrodes (E) direct the microtubulesat a bifurcation, directing cargo-loaded microtubules into the sample chamber for further analysis suchas PCR amplification or sequencing. (Image taken, with kind permission from Springer Science+Busi-ness Media, from Jia et al. [1].)

13.2.1.2 Kinesin buffers

• Lysis buffer: 50-mM sodium phosphate, 300-mM NaCl, 40-mM imidazole, 5-mMβME, pH 8.0.

• Wash buffer: 50-mM sodium phosphate, 300-mM NaCl, 60-mM imidazole, 100-μMMgATP, 5-mM βME, pH 7.0.

• Elution buffer: 50-mM sodium phosphate, 300-mM NaCl, 500-mM imidazole,100-μM MgATP, 5-mM βME, 10% (w/v) sucrose, pH 7.0.

Note: For making these phosphate buffers, either combine monobasic and dibasic formsto approximate the final pH and then adjust accordingly, or use monobasic form and pHwith concentrated NaOH. pH buffers before adding βME and sucrose. βME has a lifetimeof days in these buffers, so prepare accordingly: refrigerate sucrose containing solutionsto prevent bacterial growth. One convenient approach is to make up buffers withoutATP, βME, or sucrose, and add these just before using.

13.2.1.3 Kinesin equipment

• 2L Erlenmeyer flask (preferably baffled)

• Temperature-controlled incubator with orbital shaker

• Sonicator for lysing bacterial cells (Branson Ultrasonics)

• AKTA-FPLC chromatography system (or equivalent)

• Ni-NTA agarose chromatography column (5 ml)

• SDS-PAGE gel electrophoresis equipment

13.2.2 Tubulin purification materials

13.2.2.1 Tubulin reagents

• Cow brains from freshly slaughtered cows (obtain from local slaughterhouse)

• Leupeptin hydrochloride (Sigma-Aldrich, Cat # L9783)

• Trypsin inhibitor (Sigma-Aldrich, Cat # T9253)

• Pepstatin A (Sigma-Aldrich, Cat # P5318)

• Aprotinin (Sigma-Aldrich, Cat # A6103)

• Nα-p-Tosyl-L-arginine methyl ester hydrochloride (TAME) (Sigma-Aldrich, Cat #T4626)

• Tosyl-L-phenylalanyl-chloromethane (TCPK) (Sigma-Aldrich, Cat # T4376)

• Ethylene glycol-bis(2-aminoethylether)-N,N,N‘,N‘-tetraacetic acid (EGTA)(Sigma-Aldrich, Cat # E3889)

• Ethylenediaminetetraacetic acid (EDTA) (Sigma-Aldrich, Cat # E9884)

• L-Glutamic acid monosodium salt monohydrate (sodium glutamate)(Sigma-Aldrich, Cat # G2834)

• L-Glutamic acid potassium salt monohydrate (potassium glutamate)(Sigma-Aldrich, Cat # G1501)

• Piperazine-N,N’-bis(2-ethanesulfonic acid) (PIPES) (Sigma-Aldrich, Cat # P6757)


314

• N-(2-Hydroxyethyl)piperazine-N’-(2-ethanesulfonic acid) (HEPES) (Sigma-Aldrich,Cat # H3375)

• Glycerol (Sigma-Aldrich, Cat # G5516)

• DMSO anhydrous (Sigma Aldrich, Cat # 276855)

• 2’-Guanosine 5’-triphosphate, sodium salt (GTP) (Jena Bioscience, Cat # NU-1012)

• Paclitaxel (Taxol) (Sigma Aldrich, Cat # T7402) (Note: Required for stabilizingmicrotubules. Dissolve at 1 mM in DMSO for use and store frozen aliquots.Taxol inhibits cell division and DMSO aids absorption into skin, so avoid contactwith skin.)

• 5-(and-6)-carboxytetramethylrhodamine, succinimidyl ester (Rhodamine-NHS)(Invitrogen, Cat # C1171)

13.2.2.2 Tubulin buffers and solutions

• PBS buffer—20-mM sodium phosphate, 150-mM NaCl, pH 7.4

• Glutamate buffer—1.0M sodium glutamate, 2-mM EGTA, 0.1-mM EDTA, 2-mMMgCl2, pH 6.6

• BRB80 buffer—80-mM K-PIPES, 1-mM EGTA, 1-mM MgCl2, pH 6.85. (Can makewith mixture of PIPES acid and salt forms or use all acid form and pH with KOH.PIPES is only minimally soluble below pH 6 and will dissolve as base is added.)

• 5x BRB80 buffer—400-mM K-PIPES, 5-mM EGTA, 5-mM MgCl2, pH 6.85

• HEPES/40% glycerol (labeling buffer)—0.1 M HEPES, 1-mM MgCl2, 1-mM EGTA,40% glycerol (v/v), pH 8.6

• Quench solution—0.5 M potassium glutamate, pH 8.6

• Depolymerization solution—50-mM potassium glutatmate, 0.5-mM MgCl2, pH 7.0

13.2.2.3 Tubulin equipment

• Waring blender

• Temperature-controlled 37°C water bath

• Beckman Ultracentrifuge, Type 19 and 50.2 Ti rotors, and centrifuge tubes/bottles

• Tissue tearor

• Beckman Airfuge

• UV-Vis Spectrophotometer

13.2.3 Microtubule gliding assay materials

• Fisher’s Finest plain glass microscope slides (Fisher Scientific, Cat # 12-544-1)

• Corning 18 x18-mm coverslips (Fisher Scientific, Cat # 12-519A) (Note: Corningcoverglass works best for gliding assay compared to other coverglasses. Differentsizes can be used)

• Double-stick tape

• Casein from bovine milk (Sigma-Aldrich, Cat # C3400) (Note: This is an importantreagent that is used to optimize surfaces for functional kinesin adsorption. Caseinshould be dissolved in BRB80 at ~20 mg/ml and stored at –20°C. Dissolve 2.5g

13.2 Materials

315

casein in 25-ml BRB80 buffer in 50-ml Falcon tube and rock for 4 hours to over-night to dissolve (keep at 4°C if > 4 hours). Next, spin in centrifuge (10 minutes at>100,000 x g is good, but less is acceptable) to pellet insoluble material. Finally, fil-ter using 0.2-μm syringe filters. You will need to go through a number of filters asthey tend to clog. Assess concentration by UV absorbance (1 mg/ml = 1 A280), if > 20mg/ml, then dilute, aliquot, and store at –20°C.)

• D-glucose (Sigma-Aldrich, Cat # G7528)

• Glucose oxidase (Sigma-Aldrich, Cat # G2133)

• Catalase (Sigma-Aldrich, Cat # C515)

Note: Glucose, glucose oxidase, and catalase are parts of an antifade oxygen scavengingsystem. It is convenient to make up stock solutions in BRB80 at 2M, 2 mg/ml, and 0.8mg/ml, respectively, store 10-μl aliquots at –20°C, and thaw and dilute 100-fold intomotility solution for motility experiments.

• Fluorescence microscope with 60x or 100x objective.

Note: Generally a CCD camera, monitor, and recording device such as a VCR or com-puter is necessary for optimal visualization and analysis. While expensive enhancedCCD cameras such as the Roper Cascade 512 work well, we have also had success with aGenwac 902H CCD camera, which can be found online for less than $400.)

13.2.4 Microfabrication materials

13.2.4.1 Microfabrication reagents and supplies

• Borosilicate glass substrates (1-mm thick)

• ProSciTech 50 × 50-mm coverslips

• 996-kDa poly(methyl methacrylate) (PMMA)

• Chlorobenzene

• Acetone

• Isopropyl alcohol (IPA)

• Sulfuric acid (H2SO4)

• Hydrogen peroxide (H2O2)

• Piranha solution (H2SO4:H2O2 = 4:1)

• 1811 photoresist

• Tetramethylammonium hydroxide (TMAH) developer

• SU-8 photoresist

• SU-8 developer

• Chrome and gold for deposition

• Gold for electroplating

• Chrome etch solution

• Gold etch solution

• Hydrogen fluoride (HF)

• Ammonium fluoride (NH4F)

• Buffered oxide etch (BOE) (NH4F/HF 10:1)


316

Note: Piranha solution sulfuric acid, hydrogen fluoride, ammonium fluoride, Cr and Auetch solutions and BOE are extremely corrosive and precautions should be taken toprevent any contact with skin or clothing. Always use proper safety procedures whenusing these chemicals.

13.2.4.2 Microfabrication equipment

• Karl Suss MA-55 aligner

• Photoresist spinner

• Hotplate

• Chemical wet bench

• DC power supply (for Au electroplating)

• Resistive filament thermal evaporator

• RIE equipment (for oxygen plasma)

• Warner 100 hydraulic press laminator

13.3 Methods

13.3.1 Kinesin expression and purification

A number of kinesins have been investigated from many different organisms, but mostof the biophysical investigations to date and virtually all of the microfluidics researchusing kinesins have employed conventional kinesin (also called Kinesin-1 [5]). Ourinvestigations have employed full-length Drosophila conventional kinesin that containsa hexaHis-tag on its C-terminus for purification [6, 7]. The full-length gene can be trun-cated to remove the tail domain and the tail replaced with a biotinylation sequence,green fluorescent protein, or other sequences. In our work we also employ a headlesskinesin construct that retains the rod and tail domains and hexaHis tag, but lacks itsmotor domain [6, 8]. Below, we provide a detailed protocol for expressing this kinesin inbacteria, purification by Ni column chromatography, and freezing for long term storage.

This protocol is for full-length Drosophila conventional kinesin in a pET plasmidexpressed in BL21(DE3) E. coli cells, though it will work for other hexaHis-taggedkinesins. It generally follows protocols described in Hancock et al. [6] and in Coy et al.[7]. Related protocols can be found in a book of kinesin protocols [9] and on the Kinesinhome page (http://www.cellbio.duke.edu/kinesin//).

The amino acid sequence of the protein and the nucleotide sequence of the entireplasmid (pPK113) are online (entry AF053733 at http://www.ncbi.nlm.nih.govpubmed/). Furthermore, this plasmid is available to academic researchers from theplasmid depository Addgene (http://www.addgene.org). The method for preparing andpurifying these proteins is described below:

1. Prepare LB media and autoclave to sterilize. After the media cools to roomtemperature, add 50 mg of ampicillin sodium salt (final concentration 100 mg/L).

2. Pipette 5 ml of LB+Amp into a 15-ml Falcon tube and inoculate it with a colony ofplasmid-containing BL21(DE3) cells. Incubate overnight (16 hours) at 37°C whileshaking at 250 rpm.

13.3 Methods

317

3. Transfer the 5-ml overnight culture to 500-ml LB medium in a 2L Erlenmeyer flaskand incubate at 37 °C while shaking at 250 rpm until the culture grows to an opticaldensity at 600 nm at of 0.5 to 1.0. This typically requires 3 to 6 hours of growth.

4. Cool the culture to room temperature on ice and add 45 mg of IPTG (final conc. 0.5mM) to induce protein expression. Incubate at 20°C for 3 hours while shaking at 250rpm to express protein. (Note: Expression at this temperature improves solubility ofthe expressed protein.)

5. Transfer culture to centrifuge bottles and harvest cells by centrifuging for 10minutes at 6,000 x g. Resuspend pellet in 25 ml of lysis buffer. If desired, cells canbe flash-frozen in liquid nitrogen and stored at -80°C at this point for laterpurification.

6. Thaw cells in 37°C water bath if necessary. Add 0.1-mM ATP and 1-mM PMSF to thesolution and lyse the cells by sonicating the cells on ice. Use to 4 to 6 pulses of 20seconds each on duty cycle 5 to completely lyse the cells. Cell lysis can also be carriedout by French press [6]. (Note: one indicator of lysis is that the solution goes fromcloudy to clear due to fragmenting the light scattering bacterial cells. Another test isto run a gel of the pellet and supernatant from step 7—incomplete lysis results in alarge portion of the expressed protein in the pellet.)

7. To remove cell debris and insoluble components, centrifuge the cell lysate at 100,000× g for 30 minutes at 4°C. The supernatant (clarified lysate) contains the solublerecombinant protein along with bacterial proteins.

8. While the cell lysate is spinning, equilibrate the Ni-NTA column by flowing through10x column volumes (50 ml) of lysis buffer at a flow rate of 5 ml/min.

9. Apply clarified lysate to the column at a flow rate of 2 ml/min to bind hexaHis-tagged protein to the column. After loading, wash column with 10x column volume(50 ml) wash buffer at a flow rate of 5 ml/min to remove unwanted bacterialproteins. The UV absorbance readout on the chromatography system (280 nm)should fall from a peak during loading to a flat baseline at the end of the wash. (Note:if greater purity of final eluted protein is desired, imidazole in wash buffer can beraised to 80 or 100 mM.)

10. Elute the kinesin by flowing elution buffer over the column at a flow rate of 1ml/min. Collect the elution in 0.5-ml fractions. Monitor elution by UV absorbance—because imidazole has an absorbance at 280 nm, there will be an absorbance jumpdue to the high imidazole concentration in the elution buffer, but a peak or at least ashoulder from the eluted protein should be observable before the absorbanceplateaus. The fractions corresponding to the peak absorbance will contain thepurified protein. The peak fractions can also be determined and the proteinconcentration quantified by a UV-Vis spectrophotometer or by running an SDS-PAGE gel of the elution fractions. Typical yields are 1 to 2 ml at 100 µg/mlconcentration for full-length kinesin, though they can be significantly higher fortruncated constructs. (Note: microtubule gliding assays using 5x-diluted elutionfractions can also be run to identify the peak fractions.)

11. The purified protein can be flash-frozen in liquid nitrogen and stored at –80°C. (Note:the 10 % sucrose in the elution buffer acts as a cryoprotectant.)


318

Notes: Kinesin frozen at –80°C is stable for years, but samples stored at 4°C only last fordays and samples at room temperatures lose activity over hours. Low-concentrationsamples can be stabilized somewhat by adding 0.5-mg/ml casein or BSA—part of theactivity loss in samples with low kinesin concentrations is due protein adsorption to thesides of the tubes, which is minimized by adding these inert proteins.

13.3.2 Tubulin purification and labeling

Microtubules, which are 25 nm in diameter and can be tens of microns in length, aremade up of α-β heterodimers of the protein tubulin [10]. There are no established tech-niques for bacterial expression of tubulin, so the standard approach is to purify tubulinfrom cow or pig brains. Microtubules are then polymerized from this purified tubulinand stabilized against depolymerization using the drug Taxol. There are publishedprotocols for large scale tubulin purification, including those by Williams and Lee[11] and Castoldi and Popov [12], as well as online protocols from the Salmon Lab(http://www.bio.unc.edu/Faculty/Salmon/lab/protocolsporcinetubulin.html) andMitchison Lab (http://mitchison.med.harvard.edu/protocols/tubprep.html) Web sites.Here, we describe an adapted protocol for isolating bovine brain tubulin that has workedreproducibly in our laboratory. The procedure involves polymerizing microtubules,pelleting them by centrifugation to remove unwanted soluble proteins, and thendepolymerizing and centrifuging to remove unwanted insoluble proteins. The key hur-dle in isolating pure tubulin is removing unwanted microtubule associated proteins(MAPs). Earlier protocols used a final phosphocellulose cation exchange column step toremove these positively charged MAPs from the negatively charged tubulin, while thisprotocol uses high concentrations of glutamate in the buffers to reduce the affinity ofthe MAPs for the microtubules.

Note that tubulin is commercially available from Cytoskeleton, Inc. (www.cytoskeleton.com). For small-scale studies, this is a reasonable source and is cost-effec-tive. However, for full control over the materials, to enable the ability to carry outdiverse modifications of the protein, and for applications where the large materialrequirements make purchasing it prohibitively expensive, it is desirable to purify thetubulin from the source.

13.3.2.1 Protocol for tubulin purification

The procedure starts with harvesting cow brains, homogenizing them to break openthe cells, and clarifying the homogenate by centrifugation to obtain a solution ofsoluble intracellular proteins. This is followed by two cycles of polymerization/centrifugation/depolymerization/centrifugation to purify tubulin away from unwantedcellular proteins and MAPs. When carried out successfully, the protocol yields roughly500 mg of pure tubulin at >99% purity, though yields will vary.

1. Prepare 2L each of PBS buffer and glutamate buffer and chill them to 4°C.

2. Obtain two cow brains from freshly slaughtered cows. Transport the brainssubmerged in ice-cold PBS buffer in an ice-filled cooler. (Note: It is critical to processthe brains within ~1 hour of slaughter. Delays in processing reduce yield due todegradation of tubulin.)

13.3 Methods

319

3. Dissect the brains on ice, discarding brain stem, cerebellum, and corpus collosum.Strip meninges (tough filaments on outside of brain) and remove blood clots, thenrinse dissected cerebrums with ice-cold PBS. Work quickly and try to balance speedwith thoroughness of cleaning—90% is a good rule here.

4. Prepare protease inhibitor cocktail by dissolving the protease inhibitors listed inTable 13.1 in 1 ml of their respective solvents.

5. Weigh cerebrums (in preweighed beaker) and transfer to a chilled Waring blender.Add 50% v/w of chilled glutamate buffer (for example: add 350 ml for 700g ofbrains). Add protease cocktails and 1-mM ATP, 0.25-mM GTP, 4-μM DTT, 0.1% βME.

6. Homogenize the brains (blending in 4×10-second high-speed pulses, with 15-secondpauses between pulses to minimize heating), and distribute the homogenate into250 ml centrifuge bottles.

7. Spin the homogenate at 50,000 x g (i.e., Beckman type 19 rotor, 19,000 rpm [53,900× g, k = 951]) for 60 minutes at 4°C to clarify the solution. Pour the supernatantsgently into a 1L graduated cylinder without disturbing the soft pellets. Discardpellets.

8. To polymerize tubulin add prewarmed (37°C) glycerol to the supernatant to a finalconcentration of 33% (v/v) glycerol, and add 0.5-mM GTP, 1.5-mM ATP eachdissolved in ~1 ml of buffer. Mix well by inverting (cover with Parafilm and hold inplace with your hand) and transfer the contents to a 500-ml stainless steel beaker.Incubate Parafilm-covered beaker in 37°C water bath for 1 hour to polymerizemicrotubules. (Note: Solution should become more viscous and take on a subtleopalescence if polymerization is proceeding.)

9. In the meantime, warm the centrifuge and rotor to 37°C. After polymerization,transfer microtubule solution to prewarmed centrifuge tubes and centrifuge at 37°Cat 300,000 × g for 30 minutes to pellet microtubules. We used a Beckman 50.2 Tirotor at 50,000 rpm [302,000 × g, k = 133]).

10. Discard supernatant containing unwanted proteins, weigh the pellets, andresuspend pellets in ice-cold glutamate buffer. Add 1 ml of cold buffer to eachcentrifuge tube and physically dislodge the pellet from the tube using a spatula.Pool the pellets in a 250-mL glass beaker on ice and add ice-cold buffer. Choosevolume for resuspension to be roughly 1/3 of polymerization volume to maintainhigh tubulin concentrations. To depolymerize microtubules, homogenize pelletswith tissue tearor and incubate 30 minutes on ice with additional tissue tearorhomogenization during the incubation.


320

Table 13.1 Protease Inhibitors

Component Final Solvent

Leupeptin 5 mg/mL DI waterTrypsin inhibitor 10 mg/mL DI waterPepstatin 1 mg/mL DMSOAprotinin 1 mg/mL DMSOTAME 1 mg/mL DMSOTPCK 1 mg/mL DMSOPMSF 100 mM Ethanol

11. In the meantime, chill the rotor and centrifuge to 4°C. Following incubation, clarifythe depolymerized tubulin by centrifuging at 300,000 × g for 30 minutes at 4°C. Savesupernatant and discard pellets.

12. Repeat tubulin cycling steps 8 to 11, with the exception that the pellets areresuspended in cold BRB80 buffer instead of glutamate buffer in step 10. This“twice-cycled tubulin,” cycled in high (glutamate) buffer to remove microtubuleassociated proteins, is equivalent to “PC tubulin” obtained by passing the solutionover a phosphocellulose column.

13. The tubulin concentration in the final supernatant can be quantified by UVabsorbance at 280 nm (1 A280 = 1 mg/ml tubulin). Aliquot tubulin into cryotubes,flash-freeze on liquid nitrogen, and store at –80°C. A helpful rule to remember is that1 mg/ml of the 100-kD tubulin dimer equals 10-μM tubulin.

13.3.2.2 Protocol for labeling tubulin with rhodamine

The following protocol describes labeling of pure tubulin (obtained in the previous step)by rhodamine succinimidyl ester (rhodamine-NHS), a reactive form of the rhodaminefluorophore. There are a range of fluorophores that can be purchased in this reactiveform as well as biotin and other compounds. These can be purchased from MolecularProbes/Invitrogen or other sources, and this protocol is suitable for microtubule labelingusing these compounds as well. This protocol is adapted from a widely used protocol inthe literature [13].

1. Warm the rotor (i.e., Beckman 50.2 Ti) and centrifuge to 37°C. Thaw a 100-mgaliquot of purified tubulin at 37°C (can be scaled up or down). Add prewarmedglycerol to 33% final v/v, along with 1-mM GTP and 5-mM MgCl2 and polymerize for30 minutes in a 37°C water bath.

2. Spin the polymerized solution at 300,000 x g for 30 minutes to pellet microtubules.Discard the supernatant and resuspend the pellet in 3.5-ml prewarmed HEPES/40%glycerol (with 1-mM GTP and 5-mM MgCl2), and maintain the solution at 37°C.(Note: Make sure the microtubules are well suspended by pipetting up and downrepeatedly.) Quickly measure the tubulin concentration using UV absorbance.

3. Just prior to labeling, dissolve the rhodamine-NHS dye (5-(and-6)- carbo-xytetramethylrhodamine, succinimidyl ester, Invitrogen) in DMSO to a finalconcentration of 50 mM. Based on the microtubule concentration add the dye in theratio of 20 dye molecules per tubulin heterodimer and incubate at 37°C for 15minutes. (Notes: (1) It is advisable to briefly centrifuge the dissolved dye solution(i.e., 1 minute in any bench top centrifuge) to remove any undissolved dye crystalsthat will later pellet with the microtubules. (2) HEPES buffer at pH 8.6 is used becauseNHS reactive groups label ionized (NH3

+) amino groups and this is closer to the pKafor the lysine side chains that are labeled. We have, however, had success withlabeling in BRB80 at pH 6.8, so this is an option for optimizing labeling.)

4. Stop reaction by adding K-glutamate to a final concentration of 50 mM. Centrifugethe solution at 300,000 x g for 30 minutes at 37°C to pellet microtubules away fromunincorporated dye.

5. Discard supernatant and resuspend the pellet in 2 ml of cold depolymerizationbuffer. Incubate on ice for 30 minutes, with occasional gentle vortexing todepolymerize.

13.3 Methods

321

6. Centrifuge the solution in a Beckman Airfuge at 30 psi for 10 minutes to removeinsoluble material and protein aggregates. Remove the supernatant and add 20%volume of 5x BRB80 buffer.

7. Polymerize microtubules by adding 1-mM GTP, 5-mM MgCl2, and 5% DMSO andincubating at 37°C for 30 minutes. Centrifuge the solution 5 minutes in the Airfuge,remove the supernatant, resuspend the microtubules in 1.5 ml of cold BRB80, anddepolymerize on ice for 30 minutes.

8. Centrifuge the solution one last time in the Airfuge and harvest supernatantcontaining the labeled tubulin. To quantify the yield and the dye:tubulin ratio,measure the absorbance in a UV/Vis spectrophotometer at 280 and 555 nm. The dyehas absorbance at both 280 and 555 nm (absorption coefficient 555 = 49,291 M–1cm–1

and 280/ 555 = 0.2476) and tubulin has an absorbance at 280 nm ( 280 = 101,900M–1cm–1). The dye concentration is obtained at 555 nm, and the tubulinconcentration is obtained at 280 nm after subtracting out the dye absorbance at thiswavelength. A 1:1 dye:tubulin ratio is a good target, though even minimally labeledtubulin is usually useful. Aliquot the labeled tubulin, flash-freeze in liquid nitrogen,and store at –80 °C.

13.3.2.3 Cycling tubulin and mixing labeled unlabeled tubulin for optimalfunctionality

For visualizing microtubules by epifluorescence microscopy, we have found that theoptimum dye:tubulin ratio is 1:5. This ratio gives microtubules that are bright enough tovisualize, and the dye has no apparent effect on microtubule or kinesin function. Theapproach is to combine labeled and unlabeled tubulin at specified ratios and to aliquot,freeze and store this tubulin for use in motility experiments. We have observed thatmicrotubules containing 1:1 dye:tubulin ratios have different polymerization character-istics, so caution should be used when working with highly labeled microtubules, butthe optimum dye:tubulin ratio may vary for different applications, and can be modi-fied. Furthermore, multifunctional microtubules that contain tubulin labeled withcombinations of fluorophores or with biotin can be assembled (i.e., biotinylated,rhodamine-labeled microtubules). When making up tubulin solutions, unlabeledtubulin is typically cycled, quantified, and then combined with labeled tubulin. Thefollowing cycling protocol ensures that all of the tubulin is active and removes any deadprotein.

1. Thaw a 50-mg aliquot of tubulin in a 37°C water bath. Add 1-mM GTP, 5-mM MgCl2

and 5% DMSO and incubate at 37°C for 30 minutes to polymerize.

2. Centrifuge the polymerized microtubules at 300,000 × g for 30 minutes (i.e., 50,000rpm in a 50.2Ti rotor) at 37°C. Discard supernatant, resuspend the pellet in 1.5 ml ofcold BRB80 buffer, and depolymerize on ice for 30 minutes.

3. Centrifuge the solution in an Airfuge for 10 minutes and harvest the supernatant.Measure the concentration of cycled tubulin using UV absorbance.

For normal rhodamine-labeled microtubules, the cycled tubulin and rhodamine-labeled tubulin are mixed to yield a dye:tubulin ratio of 1:5, and the solution is dilutedusing BRB80 buffer to a final concentration of 4 mg/ml. The solution is aliquotted into

10-μl aliquots, flash-frozen in liquid nitrogen, and stored at −80°C for later use.


322

13.3.2.4 Polymerizing and stabilizing microtubules for microtubule gliding assays

To polymerize microtubules for kinesin motility experiments, thaw a 10-μl aliquot ofrhodamine tubulin and add the components listed in Table 13.2.

Mix by briefly flicking tube or pipetting, and incubate the solution at 37°C for 20minutes to polymerize the microtubules. Meanwhile, make a 1-ml solution of BRB80 +10-μM Taxol (BRB80T) by adding 10 μl of 1-mM Taxol in DMSO to 990-μl BRB80 andrapidly vortexing to disperse the Taxol. Following polymerization, add the entiremicrotubule solution to the BRB80T solution and disperse by inverting the tube repeat-edly. This results in the Taxol binding to the microtubules and stabilizing in theirpolymer form. These microtubules (called MT100 for 100-fold diluted microtubules) arestable for a week on the bench at room temperature. However, placing the tube on ice fora few minutes will depolymerize the microtubules. This microtubule solution can beflash-frozen on liquid nitrogen and stored at –80°C; thaw rapidly (warm hands workwell) to prevent depolymerization.

13.3.3 Standard protocol for the microtubule gliding assay

The microtubule gliding assay involves immobilizing kinesin motors on a glass cover-slip, introducing fluorescent microtubules in the presence of ATP, and observing theirkinesin-driven transport. Versions of this assay are published in [14, 15] and online(http://www.cellbio.duke.edu/kinesin//), and there are many potential modificationsthat can be made. This assay is included here because it provides the foundational exper-iments for achieving kinesin-driven transport in microfluidic devices. It is also routinelyused as a control experiment to assess the function of motors, microtubules, and otherreagents. The protocol for the microtubule gliding assay are as follows:

1. Prepare a flow cell using a glass microscope slide and 1½ coverslip, using double-sticktape as a spacer. The depth of the cell will be ~100 μm, and the width can be set bywhere the tape is placed.

2. Prepare 1 ml each of three solutions. BRB80CS is a casein-blocking solution con-sisting of 0.5-mg/ml filtered casein in BRB80. BRB80CA, made of 0.2-mg/ml caseinand 1-mM ATP in BRB80 is used for motor dilutions. BRB80CT, made of 0.2-mg/mlcasein and 10-μM paclitaxel in BRB80 is used for microtubule solutions.

2. Block the surface by flowing 50 μl of BRB80CS into the flow cell and incubating for 5minutes. By carefully pipetting the solution into the edge of the experimentalchamber, capillary action will draw the solution into the flow cell, filling it. Thecasein adsorbs to the glass surface and prevents denaturation of the motors on thesurface.

13.3 Methods

323

Table 13.2 Microtubule Polymerization Recipe

Vol. Reagent (Stack Conc.) Final Conc.

10 μl Rho-tubulin (4 mg/ml, 1:5rho:tubulin)

32 μM

0.5 μl MgCl2 (100 mM) 4 mM0.5 μl MgGTP (25 mM) 1 mM0.6 μl DMSO 5%0.9 μl BRB80Total 12.5 μl

3. Dilute the purified kinesin to 1 μg/ml in BRB80CA, and introduce 50 μl of this motorsolution into the flow cell. Solution exchange is achieved by introducing thesolution into one end of the flow cell using a pipette while simultaneously wickingthe solution out of the other end of the flow cell using a Kim Wipe, filter paper, orother absorbent material. Incubate for 5 minutes to allow motor binding to the glasssurface. (Note: Useful starting motor concentrations are from 0.05 to 5 μg/ml, andcan be optimized empirically.)

4. For making motility solution, use stock solutions described in the “Materials”section with the recipe listed in Table 13.3.

This motility solution contains microtubules, MgATP, Taxol to stabilize the micro-tubules, casein to maintain an adsorbed layer on the surface, and an oxygenscavenging system consisting of glucose, glucose oxidase, catalase, andβ-mercaptoethanol to prevent photobleaching. Flush the flow cell with 50 μl of thismotility solution.

5. Observe microtubule movement using epifluorescence microscopy. It is possible toobserve microtubule movements by eye, but optimal performance is achieved byusing a CCD camera attached to the microscope.

13.3.4 Design considerations for integrating motor proteins intomicrofluidic devices

To successfully integrate motor proteins in microfluidic devices, the following issuesneed to be considered in the design of the device.

1. Compatibility of materials. The chosen materials should be transparent and supportkinesin-driven motion. Some traditional photoresists such as 1811 areautofluorescent and as such they hinder the visualization of fluorescentmicrotubules. We have assessed the activity of kinesin motors on a range of surfacesand found that motors retain their function when adsorbed to a number of differenthydrophilic surfaces including glass, gold and chrome electrodes, plasma-treatedsilicon, and plasma-treated SU-8. However, motors lose their function and/ordenature on hydrophobic surfaces such as hydrophobic silanes and untreated SU-8[16, 17]. In addition, some materials such PDMS cannot be used with fluorescentmicrotubules, as microtubules are photobleached and depolymerize under themicroscope even in the presence of an antifade system. This behavior has beenattributed to the high oxygen permeability of PDMS, which overwhelms the oxygenscavenging activity of the antifade system [18, 19].


324

Table 13.3 Mobility Solution Recipe

Vol. Reagent (Stock Conc.) Final Conc.

85 μl BRB80CT1 μl MgATP (100 mM) 1 mM1 μl D-glucose (2 M) 20 mM1 μl glucose oxidase (2 mg/ml) 20 mg/ml1 μl catalase (0.8 mg/ml) 8 mg/ml0.5 μl β-mercaptoethanol 0.5%10 μl MT100 (0.32 µM) ~32 nM100 μl

2. Controlling microtubule movement. In the standard gliding assay, microtubulesmove in all directions across the surface, and the directionality is primarilydetermined by the orientation of the microtubules when they land on the surface. Anumber of groups have shown that microfabricated surface features such as wallsand channels can be used to control the trajectory of kinesin-driven microtubules[17–22] (see Figure 13.2). One design feature that comes out of these studies is thatcontrolling the direction of microtubule movement requires channels with widthsless than 10 µm (the average length of microtubules polymerized in vitro). Due tothe high flexural rigidity of microtubules, confinement in this size range forces themicrotubules to follow the microchannel trajectory. Smaller channels providebetter confinement, but control of fluid flow and solution exchange is increasinglychallenging as the channel width is decreased.

3. Fluid flow and solution exchange. When enclosed microfluidic channels are used,fluidic connections from a syringe pump to the microchannels are required to enablethe sample introduction into the device. The fluidic connections should be designedfor fast solution exchange (the typical gliding assay requires two solution exchanges)and the fluidic connections should allow the sample to be mounted on a microscopestage.

4. Depletion of fuel and buffer. The surface of the typical gliding assay as describedabove is open to bulk solution so the ~100-μm layer of solution contains sufficientATP to power the motors for many hours. However, in enclosed microchannels thehigh surface-to-volume ratio can lead to depletion of ATP fuel and loss in thebuffering capacity of the solution over time [8]. Any design should thereforeconsider diffusion of these species into the channels, for instance by placingreservoirs along the length of the channel to replenish the fuel. Alternately, slowbulk flow in the channels can replenish the fuel without affecting microtubulemovement.

13.3 Methods

325

Figure 13.2 Confinement of microtubule motion using microfabricated walls. SU-8 is deposited on aglass substrate to create micron-scale walls (left). Under the fluorescence microscope with simultaneouslow-level bright field illumination, moving microtubules and the microchannel walls can be observed(right). Microtubules that collide with the walls buckle and are redirected, resulting in transport alongthe channel axis.

13.3.5 Fabricating enclosed glass channels for microtubule transport

We have developed an approach for fabricating microchannels that address all of theissues discussed in the previous section. The design consists of a three-tier hierarchicalstructure (Figure 13.3) that links functional microchannels to macroscopic fluid connec-tions. Shallow microchannels (~5-µm wide and ~ 1-µm deep) for microtubule confine-ment connect to intermediate channels (100-µm wide) that serve as reservoirs and alsoconnect to 250-µm deep macrochannels that hold fine gauge tubing for simple externalfluid connections. The micro-, intermediate-, and macroscale channels are etched in a1-mm thick borosilicate glass substrate and bonded to a cover glass using poly(methylmethacrylate) (PMMA) as an adhesive [8]. Figure 13.4 shows the processing stepsinvolved in fabrication of the hierarchical channels.

The materials used (glass and PMMA) support microtubule motility and do not inter-fere with visualization of fluorescent microtubules. PMMA is used as an adhesive tobond a glass coverslip to the glass substrate, a bonding approach that is more tolerant ofparticulates than anodic bonding or fusion bonding, and can be carried out at lowertemperatures.

The macrochannel is designed in a U-shape with two tubing inserts connected tosyringe needles to allow rapid solution exchange (See Figure 13.3). If a macrochannelwith a single tubing is employed then each solution exchange requires a significant timeto flush the solution present in the tubing and macrochannel through the narrowcross-section (1 × 5 μm) microchannel. Instead, in the U-shaped macrochannel one nee-dle can be used to introduce solution and the other needle can be blocked to directsolution into the microchannel. For solution exchange, the second needle is unblockedto flush the U-shaped macrochannel with a new solution, and then is reblocked to directthe new solution into the microchannel.


326

Figure 13.3 Hierarchical microchannel design. Macrochannels (250-μm deep) enable sample intro-duction through attached syringe needles. Intermediate channels (100-μm wide and 1-μm deep) con-nect to microchannels (5-μm wide and 1-μm deep) where microtubule motility is observed. Thebottom panel shows a completed sample including the coverglass bonded using PMMA adhesive, andtubing for sample injection. (Image adapted, with kind permission from Springer Science+BusinessMedia, from Huang et al. [8].)

13.3.5.1 Microchannel fabrication protocol

The fabrication process is described below. For detailed protocols on making masks,photolithography steps, and photoresist development, consult the microfabricationprotocols detailed in earlier chapters.

1. Clean glass substrates using acetone and isopropyl alcohol in an ultrasonic bath for10 minutes each, followed by cleaning in Piranha solution (H2SO4:H2O2 = 4:1) for 20minutes. The Piranha treatment both cleans the glass and results in a high-surfaceenergy OH-saturated surface that improves the adhesion of deposited layers.

2. Deposit a 10-nm Cr layer and a 100-nm Au layer on the cleaned glass substrate usinga resistive filament thermal evaporator. Micro- and intermediate channel designs arethen patterned on the metal layers by photolithography. First, spin on Shipley 1811photoresist, then expose the sample through a mask (we use a Karl Suss MA-55aligner), and finally develop the photoresist. (Note: The Cr/Au layer eliminatesphotoresist lifting that can be a problem for direct etching of glass with only aphotoresist mask.)

3. Etch shallow micro- and intermediate channels using a buffered oxide etch solution(NH4F/HF 10:1) at room temperature with moderate agitation. A 20-minute etchresults in 1-µm deep channels, which work well for this application. After etching,strip the photoresist, Au, and Cr layers by immersing in acetone, Au etchant, and Cretchant, respectively.

4. Etch U-shaped macroscale channels into the glass to connect the macroscopic fluidconnections to the microchannels. The glass etching uses a robust Cr/Au/SU-8

13.3 Methods

327

Figure 13.4 Schematic of fabrication process for enclosed glass microfluidic channels. Micro-and intermediate channels are first etched in the glass substrate (a), followed by etching deepmacrochannels (b). A glass coverslip is then bonded using PMMA as an adhesive (c) and tubing forfluid connections are epoxied into the macrochannels (d). See text for fabrication details. (Imagetaken, with kind permission from Springer Science+Business Media, from Huang et al. [8].)

protection layer and concentrated hydrofluoric acid (HF). First, deposit a thermallyevaporated layer of Cr (10 nm) and Au (100 nm) to achieve good adhesion to theglass surface and provide a seed layer for gold electroplating. Second, deposit 1811photoresist on this metal layer, pattern the U-shaped macrochannels in it byphotolithography using a dark field mask, and then use Cr and Au etching to exposethe glass in these regions. Third, electroplate a 1.5-μm layer of gold (Technigold 25ES, Technic Inc.) on the remaining Cr/Au. (Note: Active filtering of theelectroplating solution during the electroplating process helps to removeparticulates that can lead to pinhole defects in later steps.) Finally, spin on a 40-μmlayer of SU-8 photoresist, align to the U-shaped channels using a light field mask,and pattern the macrochannels in this SU-8 to expose the glass. The thickelectroplated Au provides good protection against concentrated HF with reduceddefect density and cost compared to evaporation or sputtering, and the SU-8photoresist provides additional protection against pinholes.

5. Etch glass for 45 minutes in 49% HF at room temperature, which results in ~300-µmdeep macrochannels. After etching, the SU-8 layer can be removed by piranhacleaning and the electroplated Au and evaporated Cr/Au layers removed by wetetching. Alternatively, the sample can be simply soaked in Cr etchant to lift off thewhole Cr/Au/Au/SU-8 layer.

6. Spin-coat a 500-nm thick PMMA layer from chlorobenzene on 48 × 50-mm cover-glass (Gold seal #1, thickness: 0.13–0.17 mm) and bake on a 180°C hot plate for 20minutes to remove the solvent and allow the PMMA layer to flow. Then, expose boththe PMMA and glass microchannel surfaces to oxygen plasma for 1 minute toincrease their adhesivity. Finally, laminate the PMMA-coated coverslip to the glasssubstrate using a Warner 100 hydraulic press laminator at 50 psi and 120°C for 10minutes, and cool to room temperature while maintaining pressure. (Note: 996kDa PMMA is chosen because it is sufficiently adhesive at convenient bondingtemperatures to bond the two layers and is sufficiently viscous to prevent filling ofthe microchannels.)

7. Insert stainless-steel tubing into the macrochannels and bond using epoxy. Connectthe other end of this tubing to syringe needles for sample introduction.

13.3.5.2 Integrating motors and microtubules into microfluidic channels

The microtubule gliding assay protocol can be integrated into microfluidic channels bysequentially flowing casein, kinesin, and microtubule solutions into the channels. How-ever, to account for the high surface-to-volume ratios in microchannels, the reagentconcentrations should be increased. The high surface-to-volume ratio also leads to gradi-ents in motor adsorption in the channels, with higher motor densities upstream andlower motor densities downstream due to depletion of motors. While this gradient canbe eliminated by flowing in large amounts of kinesin solution to saturate the surfacewith motors, the resulting high motor densities in the intermediate channels lead todense microtubule meshes forming at the macrochannel/intermediate channel junctionthat eventually clog the intermediate channels [8]. This problem can be solved using aheadless kinesin construct [6] that competes with the full-length kinesin for surfaceadsorption to the channel walls but does not interact with microtubules. By varying theratio of full-length kinesin to headless kinesin in the motor solution, an optimal surface


328

density of full-length kinesin can be obtained such that robust transport is achievedwithout dense networks of microtubules clogging the channels. The protocol for achiev-ing functional microtubule transport in microfluidic channels is described below:

1. To immobilize motors inside the microchannels, introduce a solution containing4-mg/ml casein, 7.5-μg/ml conventional kinesin and 32-μg/ml headless kinesin(KRT) in BRB80 buffer. (Note: Here we eliminated one solution exchange step bycombining casein and kinesin solutions, using a large excess of casein such thatcasein adsorption will be much faster than kinesin adsorption. Alternately, caseinsolution can be introduced first, followed by the kinesin solution.)

2. Flush in a microtubule solution containing 0.64-µM rhodamine-labeled micro-tubules, 10mM ATP, 50 µM paclitaxel and antifade reagents (0.1 M D-glucose, 0.1mg/ml glucose oxidase, 0.04mg/ml catalase, and 0 c.35M β-mercaptoethanol) inBRB80 buffer. (Note: The concentrations of ATP and antifade omponents are 10 to 20times higher than standard assay to account for depletion in the microchannels.)

3. Visualize microtubule movements in the microchannels by epifluorescence micro-scopy. To simultaneously image the microchannels and the fluorescent micro-tubules, use simultaneous bright field microscopy with low illumination levels,which results in an overlaid image of the channels and the moving microtubules (asin Figure 13.2).

13.4 Results

Successful integration of microtubule-based motility into microfluidic channels requiresboth maintaining functional activity of the kinesin and microtubule proteins, as well asdesigning and fabricating biocompatible microchannels in geometries that generateuseful transport. Here, we describe results from a standard microtubule gliding assay andthen demonstrate two microchannel applications; one that generates unidirectionalmicrotubule transport in enclosed channels and a second that accumulates dense bun-dles of uniformly aligned microtubules. These examples are from our published work[8], and for the interested reader there are a number of other published strategies fromour group and others for controlling the kinesin-driven transport of microtubules inboth open top and enclosed microchannels [1, 16, 17, 20–28].

The microtubule gliding assay is both the experimental foundation for integratingmicrotubule transport into microfluidic channels, as well as a helpful research toolfor assessing the activity of kinesin and microtubule proteins and for testing the

13.4 Results

329

Figure 13.5 Microtubules moving over a kinesin-coated surface in the microtubule gliding assay.Image acquired by fluorescence video microscopy with frames 5 seconds apart.

biocompatibility of different surface chemistries and materials. Figure 13.5 shows framesfrom a standard microtubule gliding assay, and a video is available on the Artech HouseWeb site (http://www.methodsinbioengineering.com). Immediately following theintroduction of microtubules, filaments land and move across the kinesin-covered sur-face in the range of 0.5 to 1 μm/s at room temperature. Occasionally, microtubules willdissociate from the surface and diffuse away, but when high motor densities (1-μg/mlkinesin and above) and long microtubules (> 5 μm) are used this dissociation is rare.When very low kinesin concentrations are used in the assay, transient events can beobserved in which microtubules land, are moved by individual kinesin motors, and thendiffuse away when the motor reaches the minus-end of the microtubule [14]. Investiga-tors who are exploring novel microchannel geometries, exploring new materials forsuch microchannels, or otherwise exploring this research area are advised to use themicrotubule gliding assay to fully explore experimental variables before integrating themotility into channels.

When these same motors are integrated into enclosed microfluidic channels (fabri-cated according to the protocol in Section 3.5), the channel walls guide the microtubulesand specific channel geometries can be used to achieve unidirectional microtubuletransport. As a microtubule is being transported across a kinesin-functionalized surface,the leading end of the microtubule is continually encountering new motors. In theabsence of any external forces the path the microtubule takes will tend to be fairlystraight, since on these micron dimensions microtubules are quite stiff [29]. However, ifthe tip is biased to encounter motors in one specific direction, such as when amicrotubule encounters a microfabricated barrier at an angle, the entire microtubulewill tend to follow that path. One useful feature for microtubule-based microfluidicdevices is a directional rectifier that causes microtubules to all move in the same direc-tion. Figure 13.6 shows the design and performance of one such directional rectifier.Microtubules that enter from the left are guided either to the top or the bottom channel,and they travel around the structure and exit from the left. In contrast, microtubulesthat enter from the right collide with the channel wall, are guided either up or down,and then they continue through the channels and exit to the left. Thus, this structuregenerates uniform transport directions for a population of filaments. We found that


330

Figure 13.6 A microtubule directional rectifier fabricated in enclosed glass microchannels. Schematicat left shows expected microtubule paths. Microtubules entering from the left (red trace) travel aroundthe structure and exit to the left, while microtubules that enter from the right (green trace) passthrough the structure. Image at right shows microtubules moving through the rectifier. Microtubuleoriginate from a reservoir at left side of image; the small number of microtubules to the right of thestructure demonstrate the successful redirection. (Image adapted, with kind permission from SpringerScience+Business Media, from Huang et al. [8].)

roughly 95% of the filaments are redirected in the correct orientation, and if higher tol-erances are required, multiple rectifiers can be placed in series.

Another useful structure is one that both reorients microtubules to achieve uniformtransport and acts to corral these microtubules to generate a collection of uniformlypolarized filaments. We have accomplished this by fabricating enclosed glassmicrochannels to create a ring or “roundabout” structure. The fabrication approachdetailed in Section 3.5 was used to create macrochannels for fluid introduction, interme-diate channels that act as microtubule reservoirs (see Figure 13.3), and a circularmicrochannel was created that contains input and exit ports oriented tangentially(Figure 13.7). Microtubules that enter from the reservoir at the right travel around thering, guided by the channel walls, while any filaments that change direction exitthrough one of the ports. Over time, thousands of uniformly oriented filaments can becollected that all move together through the channel. Among other uses, this systemcould serve as a starting point for cargo attachment, and if a gate were created, the fila-ments could be simultaneously directed at the chosen time into new channel.

The two microchannel structures described above are important building blockstowards realizing the device envisioned in Figure 13.1. There is clearly room for furtherdevelopments in this area, and the successful strategies presented here can be considereda starting point for future studies in this area. In other work, we and others have devel-oped approaches for attaching molecular beacons [4], magnetic nanoparticles [30, 31],quantum dots [32], and antibodies [33] to microtubules for carrying cargo. Most of thesestrategies rely on using biotinylated tubulin and using the biotin binding proteinstreptavidin as a bridge to attach the cargo. Tubulin can be biotinylated using the sameprotocol as described above for rhodamine-labeling.

13.5 Discussion of pitfalls

While there is considerable potential for incorporating microtubule motility intomicrofluidic channels, because this interdisciplinary area relies on both maintaining


331

Figure 13.7 Microtubule concentration and unidirectional transport in an enclosed glassmicrochannel ring structure. Microtubules enter from reservoirs at both the left and right and eithertravel around the ring or are guided out of the ring. At left, microtubule movements shortly after intro-duction into the channel. Microtubules move in both directions at this point. At right, an image ofconcentration ring after 90 minutes of accumulation. Microtubules are moving counterclockwisearound the ring; fluorescence intensity measurements estimate that thousands of microtubules areaccumulated. (Image adapted, with kind permission from Springer Science+Business Media, fromHuang et al. [8].)

functional proteins and fabricating useful microdevices, there are a number of points inthe process that can fail. A number of potential pitfalls and alternate strategies arediscussed here, following the same sequence used in the “Materials” and “Methods”sections.

13.5.1 Kinesin purification

The bacterial expression and column purification of kinesin has a number of potentialpitfalls, and the protocol generally requires a few runs to achieve success. In the bacterialwork, standard sterile technique is important to prevent bacterial contaminations. Inassessing protein expression, running SDS-PAGE gels of bacteria before and after induc-ing with IPTG is very helpful. As the protein is hexaHis tagged, it is possible to use com-mercial kits (Novagen is one source) for detecting His-tagged proteins. If a centrifuge isnot available, the bacterial lysate can be clarified by passing through a 0.45-micronsyringe filter (though multiple filters are generally needed due to clogging). For the chro-matography steps, other chromatography systems besides the FPLC will work fine and itis even possible to carry out batch (centrifuge-based) purifications using the Ni-NTAresin. As with expression, using SDS-PAGE gels to assay the motor concentrations in theclarified lysate, column flow through, washes, and elution is very helpful for trouble-shooting. Gels of the elution fractions will also show the relative purity and the finalconcentration of the motor sample.

Following purification, it is generally a good idea to assess motor function using themicrotubule gliding assay. Because this assay is relatively quick and is sensitive, it can beused for very low concentration samples to determine if any motor is present below thethreshold of gel detection. One mode of failure is that the purified motors are nonfunc-tional. It is important to remember that retaining kinesin functionality requires keepingsome nucleotide in the buffers (including all column buffers) at all time (10-μM MgATPminimum is a good rule of thumb). Besides the microtubule gliding assay, anothermethod for assessing motor function is to carry out a microtubule binding step in thepresence of AMP-PNP, a nucleotide analog that locks the motors onto microtubules.Motor function can be assessed by incubating motors and microtubules in 1 mMAMP-PNP, pelleting the sample in the Airfuge, and analyzing the pellet and supernatantby SDS-PAGE. Nonfunctional (nonbinding) motors will remain in the supernatant,while functional motors will pellet with the microtubules. If column-purified motors arenonfunctional, it is advised to repeat expression and test the motility of the clarifiedlysate. Although there are a number of bacterial proteins, bacteria have no microtubulemotors and if motors are functional in the bacteria, this lysate should work in glidingassays.

13.5.2 Tubulin

The tubulin purification protocol involves a number of steps that can go awry and suc-cess requires determination and persistence. On the plus side, it is reasonably easy totroubleshoot and determine which steps fail. As discussed, using fresh brains trans-ported rapidly from the slaughterhouse is important for achieving high yields. As anaside, we have noticed that yields are reduced in the summer, and a colleague RichardCyr has also reported this seasonal variation. As cows are homeotherms, the reason forthis is not at all clear, but if preps can be carried out in other seasons, that is ideal. It


332

should be noted that there have been unpublished reports of tubulin preps starting withfrozen brains, but we have no experience with this.

The first hurdle of the tubulin prep is obtaining the first microtubule pellet (follow-ing polymerization of the clarified brain homogenate). These pellets should be clearlyobservable by eye and have a somewhat opalescent quality to them. Very small pellets inthis step almost always spell doom for the rest of the purification. A second indicator ofsuccess is an increase in viscosity as the tubulin solutions polymerize. While these indi-cators are diagnostics and not fixes, they are very helpful for assessing if the prep is work-ing. One step toward avoiding failure in the tubulin prep is to ensure that the GTP isgood. Purchase from a reliable supplier and store properly.

When labeling tubulin, losses are unavoidable—a 10% yield from the thawedtubulin sample is routine. One way to avoid further losses is to maintain the tubulinconcentration above its critical concentration of ~1 mg/ml. In fact, maintaining thetubulin concentration above 5 mg/ml is recommended. Because the reactive dyes arefairly unstable, care should be taken to use dry (ideally newly purchased) DMSO for dis-solving the reactive dyes. Also, excessive labeling can reduce yields of active tubulin. Thedegree of labeling can be optimized by changing either the dye concentration or theincubation time, and different fluorophores may have different optimal labeling timesand concentrations.

For both the tubulin prep and tubulin labeling, three important points to keep inmind are (1) start with sufficient amounts of tubulin, (2) reduce volumes through theprocedures to maintain sufficiently high tubulin concentrations, and (3) work to mini-mize the time that tubulin is sitting on the bench or in an ice bucket as the protein activ-ity falls over time. While losses in tubulin activity are unavoidable, the good news is thatactive tubulin can easily be recovered by cycling such as in the tubulin labeling protocol(Section 3.2.3, steps 1–3). By polymerizing the active tubulin and pelleting it, any non-functional tubulin can be removed from the sample and, because the resuspensionvolume can be chosen, this is also a way to concentrate tubulin samples.

13.5.3 Motility assays

When working properly, the microtubule gliding assay serves as an important tool fortesting and optimizing motor and microtubule function for microchannel experiments.However, the microtubule gliding assay itself can fail for a number of reasons. First, it isimportant to start with a sufficient concentration of stabilized microtubules. The pres-ence of microtubules in the MT100 sample can be assessed by dropping a few microlitersof the solution onto a microscope slide, covering with a glass coverslip, and visualizingthe microtubules in this “squash” by epifluorescence microscopy. The next crucialreagent in the experiment is casein. Casein must be prepared as described in Materialsand stored frozen. Beware that casein solutions left on the bench for days serve as excel-lent incubators for bacteria (think spoiled milk). If there are questions with the casein, itis best to make it up new from powder. The third important component of the glidingassay is the antifade system. The enzymes scavenge dissolved oxygen in the solution andthe reducing agent further prevents free radical formation resulting from illumination ofthe fluorophores by the microscope illumination. Failure of the antifade system is fairlyeasy to assess because motility quickly ceases and the microtubules photobleach andthen fragment due to free radical attack on the proteins. This can occur at the edges of


333

the flow cell or near any bubbles in the flow cell. If this is air contact and is ruled out, themotility solution should be made from scratch using newly thawed aliquots of allantifade components. Note also that β-mercaptoethanol can oxidize over time andshould be replaced roughly every 6 months.

Motility assays can fail in two ways—either no microtubules land on the surface ormicrotubules land on the surface but they don’t move. With proper casein treatment,microtubules do not bind to the glass coverslip surface in the absence of motors, and soany bound microtubules can be attributed to kinesin activity. If, after 5 minutes nomicrotubules are seen on the surface, this generally means either that the motor concen-tration is low or the motor activity is low or absent. The first option can be tested byrepeating with a higher motor concentration. One way to enhance the signal froma small population of active motors is to repeat the motility experiment with themodification that the ATP in the motility solution is replaced with AMP-PNP. Thisnonhydrolyzable ATP analog causes motors to bind to microtubules irreversibly, so anyfilaments that do land are stuck and they accumulate over time. If no microtubule bind-ing is seen after 5 to 10 minutes in AMP-PNP, this means there is no motor activity andsuggests that the motors are completely inactive.

Another mode of failure for kinesins is irreversible binding to microtubules. If theseso-called “deadheads” are present in a motor sample, the gliding speed will generally bereduced, spiraling microtubules will be observed at high motor concentrations (due tothe front end getting stuck), and over time an increasing amount of bound but immo-bile microtubules will be observed. In very poor motor samples, it is possible to obtainonly microtubule binding and no microtubule movement; there is generally no hope ofrescuing these samples. If kinesin samples have a small number of deadheads, theactivity can be improved by simply using lower motor concentrations on the surface. Analternative step is to “clean up” the motors by incubating motors with 1-μMmicrotubules in 1-mM ATP for 10 minutes, pelleting the microtubules in the Airfuge toremove any motors that irreversibly bind, and collecting the supernatant that shouldcontain active motors. In conclusion, because achieving kinesin-driven motility inchannels rests upon reliable kinesin and microtubule samples, it is worth spending timeworking with gliding assays to fully characterize the functional activities of the motorsand microtubules.

13.5.4 Motility in microchannels

Achieving robust microtubule transport in microfluidic channels requires proper intro-duction of solutions into the channels, proper adsorption of motors to the channelwalls, and good functional activity of the kinesin motors and microtubules. Before intro-ducing motor or microtubule samples, it is often worthwhile to introduce a solution offluorescent dye or labeled protein such as rhodamine-BSA into the channels to confirmthat the channels are not occluded, that they don’t leak, and that solution can be intro-duced into all regions of the device. Pumping a solution that is very easy to visualize alsoconfirms that the bonding process can withstand the hydrostatic pressures created bythe pump or syringe used to introduce the sample.

As discussed above, motility in microchannels differs from the standard gliding assayin the high surface to volume ratios involved. For this reason, it is often important toincrease not only the motor concentrations introduced into the channels but also the


334

casein, ATP, and antifade concentrations (increases up to 10× generally don’t have anydetrimental effects). As noted in Section 3.5.2, when high concentrations of kinesinadsorb to long channels, this can prevent microtubules from entering the channelsbecause the microtubules bind and become tangled at the channel entry. Combiningthe active kinesin with headless kinesin reduces the maximal kinesin surface activityand helps to alleviate this problem. The optimum active:headless kinesin ratio is bestdetermined empirically.

Generally, troubleshooting problems with kinesin-driven motility in channels isbest achieved using the standard microtubule gliding assay. For instance, if novel sub-strates or photoresists are being used to create the microfluidic channels, flat surfacescan be created from these materials (such as by spinning the material onto coverslips),flow cells can be assembled on them, and the gliding assay can be carried out on the


335

Troubleshooting Table

Problem Explanation Potential Solutions

Kinesin purification failed. No kinesin expressed. Do miniprep and plasmid digest to confirm plasmid iscorrect. Make sure BL21(DE3) cells are used for expres-sion. Try motility assay with clarified lysate to look forany motor activity.

No kinesin purified. Confirm that Ni column is properly charged. Performmotility assay with elution peak using AMP-PNP to lookfor any microtubule binding. Confirm kinesin expression(above).

Kinesin is inactive. Confirm that there is at least 1-μM MgATP in allcolumn buffers and that expression was carried out at20°C.

Tubulin prep failed. Low yield or activity. Minimize time from cow to blender and maintain onice. Check GTP by using it to polymerize some controltubulin. Run gels of all steps to determine where pro-tein was lost.

Tubulin labeling failed. Tubulin is present but not labeled. Buy new rhodamine-NHS dye. Start with new bottle ofdry DMSO.

Low or zero tubulin yield. Decrease labeling time and/or label concentration tocontrol at ~1 label/tubulin. Spec solutions to deter-mine where protein lost.

Microtubule gliding assay fails. No microtubules on surface. Confirm that microtubules are present by visualizingsquash (see text) or focusing into solution above sur-face. Confirm kinesin concentration and use > 1 μg/ml.Replace ATP with AMP-PNP to check microtubule bind-ing.

Microtubules bind to surface but don’tmove.

Dilute motors to minimize dead head influence. Con-firm that ATP is present. Confirm that antifade is work-ing and that βME is < 6 months old (will likely see fastbleaching). May need to do new kinesin prep to obtainfunctional motors.

No motility in microchannels. Microtubules aren’t seen. Flow through dye to confirm channels are open. Visual-ize upstream to confirm microtubules aren’t trapped.Use headless kinesin at entry to minimize microtubulesticking.

Microtubules don’t move. Use higher kinesin concentrations to compensate forlarge surface area/volume. Increase concentration ofantifade components. Use standard gliding assay toconfirm that all reagents are working properly.

modified surface. We have found that for a range of surfaces, plasma treatment rendersthem hydrophilic and compatible with kinesin motors.

13.5.5 Final comments

The integration of biologically-driven transport into microengineered devices is a rela-tively new and emerging area. To date, applications have focused on microscale trans-port in channels, but there are a number of potential directions for this research. Here,we have provided the foundation protocols for harvesting and modifying kinesinmotors and microtubules as well as successful approaches for integrating these cellularproteins into engineered microdevices. These principles should provide a solid founda-tion for future investigations and will hopefully lead to functional devices based onthese technologies.

Acknowledgments

The authors thank members of the Hancock and Jackson labs, particularly Lili Jia, SamiraMoorjani, Yangrong Zhang, Gayatri Muthukrishnan, and Zach Donhauser. This projectwas funded by the Penn State Center for Nanoscale Science (NSF MRSEC DMR0213623)and by an NSF Biophotonics Grant (0323024) to W.O.H. and T.N.J. funded jointly by theNSF and NIH/NIBIB.

References

[1] Jia, L., et al., “Microscale transport and sorting by kinesin molecular motors,” BiomedicalMicrodevices. Vol. 6, No. 1, 2004, pp. 67–74.

[2] Hancock, W.O., “Protein-based nanotechnology: Kinesin-microtubule driven systems forbioanalytical applications,” in Nanodevices for Life Sciences, pp. 241–271, C. Kumar, (ed.),Wiley-VCH: Weinheim, Germany, 2006.

[3] Hess, H., “Materials science. Toward devices powered by biomolecular motors,” Science. Vol. 312,No. 5775, 2006, pp. 860–861.

[4] Raab, M., and W.O. Hancock, “Transport and detection of unlabeled nucleotide targets bymicrotubules functionalized with molecular beacons,” Biotechnol Bioeng. Vol. 99, No. 4, 2008, pp.764–773.

[5] Lawrence, C.J., et al., “A standardized kinesin nomenclature,” J Cell Biol. Vol. 167, No. 1, 2004, pp.19–22.

[6] Hancock, W.O., and J. Howard, “Processivity of the motor protein kinesin requires two heads,” JCell Biol. Vol. 140, No. 6, 1998, pp. 1395–1405.

[7] Coy, D.L., M. Wagenbach, and J. Howard, “Kinesin takes one 8-nm step for each ATP that ithydrolyzes,” J Biol Chem. Vol. 274, No. 6, 1999, pp. 3667–3671.

[8] Huang, Y.-M., et al., “Microtubule transport, concentration and alignment in enclosedmicrofluidic channels,” Biomed Microdevices. Vol. 9, 2007, pp. 175–184.

[9] Vernos, I., Kinesin Protocols. Methods in Molecular Biology, Vol. 164. Totowa, NJ: Humana Press,2000.

[10] Desai, A., and T.J. Mitchison, “Microtubule polymerization dynamics,” Annu Rev Cell Dev Biol. Vol.13, 1997, pp. 83–117.

[11] Williams, R.C., Jr., and J.C. Lee, “Preparation of tubulin from brain,” Methods Enzymol. Vol. 85 Pt B,1982, pp. 376–385.

[12] Castoldi, M., and A.V. Popov, “Purification of brain tubulin through two cycles of polymeriza-tion-depolymerization in a high-molarity buffer,” Protein Expr Purif. Vol. 32, No. 1, 2003, pp.83–88.


336

[13] Hyman, A., et al., “Preparation of modified tubulins,” Methods Enzymol. Vol. 196, 1991, pp.478–485.

[14] Howard, J., A.J. Hudspeth, and R.D. Vale, “Movement of microtubules by single kinesin mole-cules,” Nature. Vol. 342, No. 6246, 1989, pp. 154–158.

[15] Howard, J., A.J. Hunt, and S. Baek, “Assay of microtubule movement driven by single kinesin mole-cules,” Methods Cell Biol. Vol. 39, 1993, pp. 137–147.

[16] Huang, Y., et al., “Microfabricated capped channels for biomolecular motor-based transport,” IEEETransactions on Advanced Packaging. Vol. 28, No. 4, 2005, pp. 564–570.

[17] Moorjani, S., et al., “Lithographically patterned channels spatially segregate kinesin motor activityand effectively guide microtubule movements,” Nano Lett. Vol. 3, No. 5, 2003, pp. 633–637.

[18] Brunner, C., et al., “Lifetime of biomolecules in polymer-based hybrid nanodevices,”Nanotechnology Vol. 15, 2004, pp. S540–S548.

[19] Kim, T.S., et al. “Biomolecular motors as novel prime movers for microTAS: microfabrication andmaterials issues,” in 7th Int. Conf. on Micro Total Analysis Systems. 2003. Squaw Valley, CA: Trans-ducers Research Foundation.

[20] Hiratsuka, Y., et al., “Controlling the direction of kinesin-driven microtubule movements alongmicrolithographic tracks,” Biophys J. Vol. 81, No. 3, 2001, pp. 1555–1561.

[21] Hess, H., et al., “Ratchet patterns sort molecular shuttles,” Applied Physics A-Materials Science & Pro-cessing. Vol. 75, No. 2, 2002, pp. 309–313.

[22] Clemmens, J., et al., “Mechanisms of microtubule guiding on microfabricated kinesin-coatedsurfaces: Chemical and topographic surface patterns,” Langmuir. Vol. 19, No. 26, 2003, pp.10967–10974.

[23] Hess, H., et al., “Molecular shuttles operating undercover: A new photolithographic approach forthe fabrication of structured surfaces supporting directed motility,” Nano Lett. Vol. 3, No. 12, 2003,pp. 1651–1655.

[24] Clemmens, J., et al., “Motor-protein “roundabouts”: microtubules moving on kinesin–coatedtracks through engineered networks,” Lab Chip. Vol. 4, No. 2, 2004, pp. 83–86.

[25] Cheng, L., et al., “Highly Efficient Guiding of Microtubule Transport with Imprinted CYTOPNanotracks,” Small. Vol. 1, No. 4, 2005, pp. 409–414.

[26] van den Heuvel, M.G., et al., “High rectifying efficiencies of microtubule motility onKinesin-coated gold nanostructures,” Nano Lett. Vol. 5, No. 6, 2005, pp. 1117–1122.

[27] van den Heuvel, M.G., M.P. de Graaff, and C. Dekker, “Molecular sorting by electrical steering ofmicrotubules in kinesin-coated channels,” Science. Vol. 312, No. 5775, 2006, pp. 910–914.

[28] Lin, C.T., et al., “Efficient designs for powering microscale devices with nanoscale biomolecularmotors,” Small. Vol. 2, No. 2, 2006, pp. 281–287.

[29] Gittes, F., et al., “Flexural rigidity of microtubules and actin filaments measured from thermal fluc-tuations in shape,” J Cell Biol. Vol. 120, No. 4, 1993, pp. 923–934.

[30] Platt, M., et al., “Millimeter scale alignment of magnetic nanoparticle functionalized microtubulesin magnetic fields,” J Am Chem Soc. Vol. 127, No. 45, 2005, pp. 15686–15687.

[31] Hutchins, B.M., et al., “Directing transport of CoFe2O4-functionalized microtubules with mag-netic fields,” Small. Vol. 3, No. 1, 2007, pp. 126–131.

[32] Bachand, G., et al., “Assembly and transport of nanocrystal CdSe quantum dot nanocompositesusing microtubules and kinesin motor proteins,” Nano Lett. Vol. 4, No. 5, 2004, pp. 817–821.

[33] Bachand, G.D., et al., “Active capture and transport of virus particles using a biomolecularmotor-driven, nanoscale antibody sandwich assay,” Small. Vol. 2, No. 3, 2006, pp. 381–385.

Acknowledgments

337

Documents

Methods in Bioengineering: Biomicrofabrication and … · 2011-11-16 · nant kinesin, purifying and ... >100,000 x g is good, but less is acceptable) to pellet insoluble material